A cheat?!?! The rules did not mention back then that the random function could not be improved on, removing the useless repeating occurrences of the generated number. Since then, the rules have been reworded to require full execution on every nucleotide selection. At least 10 entries, Python included, improved (not cheated on) the random function. So I was just bringing Perl up to date to what seemed a perfect valid improvement shared by several entries.